Taiwan's COVID-19 Pandemic and Vaccines: The Eighteen Months That Were Bought, and the Shot That Came Too Late

On February 15, 2020, an unlicensed taxi driver died in hospital. He had never left Taiwan; he had been infected by a Taiwanese businessman he had once driven as a passenger. This was Taiwan's first COVID-19 death, and its first confirmed case of local transmission.¹

For that entire year, Taiwan reported 799 confirmed cases and 7 deaths nationwide.²

Two years later, in 2022, the same virus took 14,667 lives in Taiwan. People 65 and older accounted for 86% of them, and the median age of death was 82 — a full ten years older than the year before.³

A driver who had never left the country, and two years later, an entire cohort with a median age of 82. Between these two images lie two years — and eighteen months that one system bought for Taiwan.

Those eighteen months really were bought: through border controls, the name-based mask rationing system, and a National Health Insurance card, Taiwan kept the pandemic outside its doors for nearly a year and a half. This period was later written into a review paper in a European public health journal, in a sentence that read: "Taiwan's early containment measures might have delayed the epidemic curve, allowing time for better preparation and proactive responses."⁴

The key is in the second half of that sentence. The time was bought, but the preparation didn't keep pace with it. On May 19, 2021, the day Taiwan raised its nationwide epidemic alert to Level 3, the government's self-procured vaccines had reached a grand total of only one batch of 117,000 doses.⁵ By the time Omicron arrived in 2022, the senior citizens who needed that shot most still had their third doses lagging behind their own first and second.⁶

30-Second Overview: At eight in the morning on December 31, 2019, Taiwan's Centers for Disease Control (CDC) emailed China and the World Health Organization to verify reports of an unexplained pneumonia in Wuhan, and that same day began boarding-gate quarantine checks on flights arriving direct from Wuhan. Through border controls, the name-based mask rationing system, and a National Health Insurance card, Taiwan kept the pandemic outside for nearly eighteen months, recording only 799 confirmed cases and 7 deaths in all of 2020. The first AstraZeneca vaccines didn't arrive until March 3, 2021, and only 117,000 doses at that. By mid-July that year, government-purchased doses from the two major brands combined had reached about 2.45 million, against a total contracted volume of 15.05 million. When Omicron arrived in 2022, senior citizens' third doses still lagged behind. That year, 14,667 people died of COVID-19, with a median age of death of 82. Who bore the costs beyond death is a separate ledger that has never been fully tallied.

Eight A.M. Email, Same-Day Flight Checks

In the early morning of December 31, 2019, Lo Yi-chun, then deputy director-general and spokesperson of Taiwan's Centers for Disease Control (CDC), saw online reports of at least seven cases of atypical pneumonia in Wuhan. At eight in the morning, the CDC confirmed this with China's Center for Disease Control and Prevention and notified the World Health Organization's International Health Regulations (IHR) contact point by email. That same day, flights arriving in Taiwan direct from Wuhan were intercepted for boarding-gate quarantine inspection.⁷

Ask first, then act without waiting for an answer. That sequence would repeat itself for the next eighteen months.

That email later became material for international debate, yet the CDC has never released its full text. Officials separately added an explanation: the email had mentioned that "patients have been isolated for treatment," from which public health professionals could infer the possibility of human-to-human transmission. That clarifying line was added in April 2020 — it was not part of the text sent on December 31.⁷ In many later retellings, the two have been merged into a single statement.

On January 20, 2020, the Central Epidemic Command Center (CECC) was established under Article 17, Paragraph 2 of the Communicable Disease Control Act, opening at Level 3, with CDC Director-General Chou Chih-hao as commander. The next day, Taiwan recorded its first confirmed case — a Taiwanese businessman who had returned from Wuhan. On January 23, the command level was raised to Level 2, with Minister of Health and Welfare Chen Shih-chung taking over as commander. On February 27, it was raised again to Level 1.⁸

Source: Ministry of Health and Welfare, successive Taiwan CDC announcements

📝 Curator's Note
The "Level 3 alert" that Taiwanese people remember and the CECC's own "Level 1 activation" are two completely different systems. The former governs restrictions on people's daily lives; the latter governs the command center's own organizational tier. Once the CECC was raised to Level 1 on February 27, 2020, it was never downgraded again until it disbanded on May 1, 2023 — through the takeout-only, empty-streets summer of 2021, the command center's own tier never changed at all. Reading "Level 3 alert" as "the CECC dropped back to Level 3" is exactly where the two systems are most easily conflated. The severity people felt and the bureaucracy's own tier are two curves that never moved in sync.

At nine in the morning on February 6, 2020, the name-based mask rationing system went into effect. People could buy masks at NHI-contracted pharmacies using their National Health Insurance card: odd-numbered national ID numbers could buy on Monday, Wednesday, and Friday, even-numbered ones on Tuesday, Thursday, and Saturday, and anyone at all on Sunday.⁹ The lines Taiwanese people stood in that spring, the numbers they memorized, the pharmacy doors they stared at — that was the most physical form this system ever took.

What eventually shortened that line wasn't in the government's hands. Wu Chan-wei (吳展瑋), founder of the Tainan-based "Howdy Studio," wrote his own map of convenience-store mask stock during the mask-buying chaos; a prototype had already appeared at a Tainan hackathon in late 2019, and the g0v community gradually joined in. Then-Digital Minister Audrey Tang noticed it and reached out, coordinating data openness between g0v and the government; the government later turned pharmacies' remaining mask counts into an open API.¹⁰ The step most often left out of this chain is the first one: the map's first version came from one engineer's own hands, and only afterward did the government step in to feed it data.

That same year, the "electronic fence" system launched in early February, using cell tower signals from a person's mobile phone number to determine whether someone under home quarantine had left their quarantine location.¹¹ This system would later become a constitutional-level controversy — but that was still three years away.

By the end of 2020, Taiwan's report card read: 799 confirmed cases, 7 deaths.²

Except the claim of "eighteen months with zero community transmission" doesn't hold up — two counterexamples fall within that very period. The unlicensed taxi driver who died on February 15 was a confirmed case of local community transmission.¹ A hospital-acquired cluster at Taoyuan General Hospital in January 2021 produced 11 cases of community spread.¹² The accurate version is this: breaches did occur, and each time one did, Taiwan poured enormous resources into containing it.

March 30, 2020, Hsinchu City. The line for the name-based mask rationing system, buying with a National Health Insurance card. Photo: Tbatb / Wikimedia Commons, CC BY-SA 4.0

The April 13 Meeting That Left No Record

On April 13, 2021, the CECC held an expert meeting to discuss relaxing quarantine rules for the crews of national-flag airlines. The next day, it announced a switch to "3+11" — three days of centralized quarantine plus eleven days of self-health management — effective from April 15 onward, based on each flight's arrival time in Taiwan. On May 31, the CDC stated that this adjustment had been discussed across the CECC's divisions but that no meeting minutes had been produced. On June 12, the crew rule was rolled back again: unvaccinated crew members on long-haul flights were switched to "7+7."¹³

By late April, the outbreak in Wanhua was already spreading. The index case identified in retrospect (Case 1424) had an onset date of April 23. On May 11, seven new local cases were reported and the alert was raised to Level 2. On May 15, Taipei and New Taipei were raised to Level 3; on May 19, the whole country followed.¹⁴

The Level 3 rules were written in fine detail. All private gatherings not required for work were suspended — indoors above 5 people, outdoors above 10 — and masks had to be worn at all times outside the home. Aside from businesses essential to daily life, public order, necessary services, medical care, or government functions, all other businesses and public venues were closed, with karaoke halls, dance halls, bars, gyms, and arcades all on the list. Restaurants that couldn't enforce the rules switched to takeout only; religious pilgrimages and temple processions were suspended; and elementary and secondary schools nationwide were closed to outside access.¹⁵ The phrase "not required for work" is rarely remembered along with the rest — most people still went to their jobs as usual that summer.

At a press conference at two in the afternoon on May 22, the CECC announced two numbers: 321 new local cases, plus another 400 cases under xiàozhèng huíguī ("recalibration"), backdated across May 16 to 21. The technical cause was a bottleneck in local reporting: more than ten thousand test results were stuck at local hospitals, unable to be reported upward, and after the CECC simplified its reporting form from more than twenty fields down to eight, it processed more than fourteen thousand of them in a single batch.¹⁶ The term "recalibration" was coined on the spot at that press conference — it wasn't an existing epidemic-prevention term — and it later became everyday slang Taiwanese people use to mock anyone submitting overdue work.

May 25, 2021, Dongmen Station, Taipei MRT. A platform under the Level 3 alert. Photo: 稜光映像_Brian / Wikimedia Commons, CC BY 3.0

PTS News report from May 22, 2021: the CECC announced "recalibration" for the first time, with 321 new local cases and 400 recalibrated cases that day.

For the next year and a half, society poured its energy into the same question: who should be held responsible for that May. In August 2021, Chen Shih-chung told English-language media, "There was no three-plus-11 loophole," noting that the CECC's three-month investigation had found no link between the crew cluster and the May outbreak.¹⁷

The Control Yuan investigated for nearly a year and a half, and its report, released in August 2022, reached a far more measured conclusion than the simplified version that circulated publicly. The Control Yuan found that before deciding to implement the strengthened protocol, the CECC had already comprehensively weighed factors including the local outbreak situation at the time, the airlines' management responsibilities, crew members' PCR test results, and the physical and mental stress they were under — it had not acted arbitrarily without evaluation. The criticism centered on the CECC's failure to fully explain its decision to the public, and its failure to proactively communicate when media and legislators raised questions. No violation of law was found, and no corrective measure was issued. The Control Yuan simultaneously recommended amending the law to specify whether major epidemic-prevention decisions should be documented in minutes and made public.¹⁸

That legislative recommendation reveals something: current regulations don't actually require meeting minutes to be kept. So "no record was left of the April 13 meeting" is both true and not a violation of law. It is a gap in the system, not one person's negligence. Follow the question of responsibility all the way to the end, and this is the answer you get.

That same May, Wanhua was hunted by media and the internet for a long time afterward. The shops in Wanhua bearing "tea house" signs were actually two different kinds of business that are often confused with each other: tê-toh-á ("tea-table" parlors) sold only tea and simple snacks, entirely different in nature from venues with female hostesses. The people who visited weren't necessarily socially disadvantaged — what they had in common was loneliness. A pot of tea there cost NT$200, a song on the sound system NT$10.¹⁹

June 12, 2021, outside a rapid-testing station in the Jisui area of Zhonghe District, New Taipei City. Intersection signage and barricades. Photo: Tsuna Lu (Adsa562) / Wikimedia Commons, CC BY-SA 4.0

The truly scarce thing was nowhere in this debate: on May 19, the day of the nationwide Level 3 alert, the government's self-procured vaccines still amounted to nothing more than the single batch of 117,000 doses that had arrived on March 3⁵, plus the 199,200 doses COVAX had delivered on April 4²⁰ — roughly 320,000 doses in all of Taiwan, for 23 million people.

Contracted for 15.05 Million Doses, With 2.45 Million Delivered by Mid-July

At the end of 2020, the government announced it would procure about 20 million doses of vaccine from abroad, plus advance orders for 10 million domestically made doses (5 million each from Medigen and United Biomedical (UBI)).²¹ The 20 million doses from abroad broke down as 10 million from AstraZeneca (AZ) and about 5.05 million from Moderna, plus 4.76 million ordered through the global vaccine-sharing platform COVAX, for which Taiwan had already paid a deposit in October 2020.²⁰

The first batch of 117,000 AZ doses arrived at Taoyuan International Airport at 10:21 a.m. on March 3, 2021, aboard Korean Air flight KE691.⁵

By July 15, 2021, the CECC's own announced figures for self-procured deliveries were: three AZ shipments totaling 1.303 million doses, and four Moderna shipments totaling 1.15 million doses.²² Combined, the two brands came to about 2.45 million doses — less than a fifth of the total contracted volume of 15.05 million.

Source: Ministry of Health and Welfare, Taiwan CDC / CECC Decision Hub, Legislative Yuan

Contracted doses, delivered doses, and doses that could actually go into an arm are three different numbers. The shape of the wait was exactly the gap between them.

⚠️ Contested View
The most common rebuttal to "the vaccines arrived late" is that the whole world was scrambling for doses in the first half of 2021, and the slow deliveries were a supply-chain problem, not a choice Taiwan made. Half of that claim holds up — global manufacturing capacity that half-year really was consumed by a handful of countries that had signed contracts early. The other half can be checked against the timeline: by July 15, 2021, Japan had donated 3.34 million doses across three shipments and the United States had donated 2.5 million doses on June 20. The 5.84 million doses donated by just these two countries already amounted to more than double the government's own procured deliveries (2.45 million).²²²³ Those donations were carved out of the very same global supply. The shortage was real. It just doesn't explain everything.

Japan ultimately donated a total of 4.204 million doses across six shipments, making it the largest single-country donor to Taiwan worldwide. The United States' 2.5 million Moderna doses arrived on June 20; Poland's 400,000 doses arrived on September 5; the Czech Republic gave 30,000 and Slovakia 10,000; Lithuania's 20,000 doses can only be confirmed through secondhand sources.²³

Source: Ministry of Foreign Affairs, Taiwan CDC / CECC Decision Hub, UDN Health

The most fiercely debated thread that summer was BNT. There were three different accounts of how the deal fell apart, each with primary sourcing, and they don't agree on why Taiwan couldn't get the vaccine.

In November 2020, TTY Biopharm's bid to become the distributor fell through. The official account is that the two sides couldn't agree on quantity and price — the government's initial procurement scale fell far short of what the manufacturer expected.²⁴

In December 2020, the government's direct negotiations with BNT reached the final stage before contract signing. The CDC's own account is that BNT asked for the words "our country" to be deleted from the Chinese-language version of the press release; the Ministry of Health and Welfare changed it to "Taiwan" and sent it out the next day, after which the negotiations quietly died.²⁵

In February 2021, Chen Shih-chung revealed that Taiwan had already agreed to purchase 5 million doses from BNT, but that the deal changed due to "external interference" just as it was about to be announced in December 2020 — he did not name who was responsible. BNT's own statement at the time read: "BioNTech is committed to helping end the pandemic around the world, and our intention to supply vaccines to Taiwan is part of that global commitment. Relevant discussions are still ongoing, and BioNTech will provide updates."²⁶

Shanghai Fosun Pharmaceutical asserted at its shareholders' meeting that it held the Taiwan distribution rights to the BNT vaccine, and that any procurement — official or private — would have to go through it; the Ministry of Health and Welfare's position was to ask the manufacturer to formally submit an import application.²⁷ Later that year, Taiwanese officials publicly accused China, for the first time, of directly obstructing the deal with BNT; China denied this throughout and said it was willing to supply vaccines to the people of Taiwan.²⁸ The only thing all three accounts agree on is that the direct negotiations at the end of 2020 really did collapse.

It was the private sector that untangled this knot. TSMC submitted its application on June 10; Foxconn founder Terry Gou's Yonglin Foundation announced its contract completed on June 12; Tzu Chi Foundation CEO Yen Po-wen went to the Taiwan Food and Drug Administration (TFDA) on June 23 to file an emergency import application and confirmed the purchase of 5 million doses on July 2; and Pegatron chairman Tung Tzu-hsien personally donated NT$500 million to help fund Tzu Chi's purchase.²⁹

Each of the three parties contracted for 5 million doses, 15 million in total, under a model of "manufactured by the original plant, packaged by the original plant, shipped direct to Taiwan" — bypassing Fosun entirely, with the CECC handling unified distribution after arrival. The first batch of 932,400 doses arrived on September 2, 2021, and the last batch on January 27, 2022; actual deliveries reached 15.2076 million doses, 207,600 more than promised.³⁰

The contract terms themselves were also debated for a long time, with two key phrases often conflated. "Sealed for thirty years" refers to the document retention period — the Ministry of Health and Welfare clarified that this was the Archives Act's rule that documents "may not be destroyed before thirty years," not a declassification deadline.³¹ Commercial confidentiality periods are a separate matter: then-Minister of Health and Welfare Hsueh Jui-yuan was explicit about this before the Legislative Yuan: "AZ and Medigen are both five years, BNT required ten years of confidentiality, and Moderna does not agree to declassification at all — never."³²

Another widely circulated claim — that "the Novavax contract has a seven-year confidentiality period" — was a misattribution. What CDC Director-General Chuang Jen-hsiang actually said was that "the Novartis contract's confidentiality clause runs seven years," referring to the 2009 H1N1 flu vaccine contract signed with Novartis — a different company, a different vaccine, and a different era from the 2021 Novavax deal.³³ The Medigen contract was ultimately made public early, on January 16, 2024, more than a year ahead of the five years originally agreed.³⁴

Five years later, what actually ended up in court was the donated money itself. On August 6, 2026, the Taichung District Prosecutors Office in Taiwan indicted seventeen people. According to the prosecution's indictment, a lawyer and a vaccine broker who claimed to be a major shareholder of Shanghai Fosun used lines such as "Foxconn and TSMC both went through Mr. Li to help procure their vaccines" to get the Tzu Chi Foundation to pay out NT$1,061,981,295. Investigators separately seized large quantities of gold and cash, and prosecutors have asked the court to confiscate more than NT$1.1 billion in criminal proceeds.³⁵ The defendants are presumed innocent until convicted.

Tzu Chi issued a statement the same day, saying that its 2021 procurement process had upheld strict principles for the use of donated funds, adhered to open-market benchmarks, and was carried out within the scope of public information, and that it had been certified by an accounting firm and reviewed by the Ministry of Health and Welfare. A lawful procurement application process and a broker's fee fraud are two different things; it is the latter that the indictment targets.

During that half-year while the shipments were still on their way, the argument at home was over something else entirely.

The Same Ruler: Medigen Passed, United Biomedical Failed Both

The legal instrument Taiwan used to approve Medigen is often called an EUA, but its actual name is "special approval for manufacture or import," under Article 48-2 of the Pharmaceutical Affairs Act — a different mechanism from what's internationally known as an EUA. The statute is explicit: the central health authority may grant special approval "to prevent, treat life-threatening or severely disabling diseases for which no appropriate drug or suitable alternative therapy exists domestically," or "to respond to an emergency or a matter that significantly affects public health." The same article's second paragraph also lays out the conditions for revocation, one of which is that "the emergency or matter significantly affecting public health has ended," and it allows authorities to order applicants to dispose of or recall unused product within a set period.³⁶

This sits at a different tier from a formal drug license. The later batch of vaccines destroyed after expiring had its legal basis written into this very article from the start.

On May 28, 2021, a TFDA expert meeting set the efficacy-evaluation benchmark for domestically made vaccines³⁷; that same day, the CDC signed contracts for 5 million doses each with Medigen and United Biomedical.³⁴ On June 10, the standard was formally announced: clinical trial reports needed at least 3,000 participants followed for at least one month after their final dose³⁸, and efficacy would be judged on two immunobridging indicators — the lower bound of the 95% confidence interval for the neutralizing-antibody geometric mean titer ratio had to reach at least 0.67 times that of the AZ control group, and the lower bound of the 95% confidence interval for the seroconversion rate had to reach at least 50%.³⁹

On July 18, Medigen's expert review meeting convened; 21 people attended, the chair did not vote, and the outcome was 3 votes to approve, 15 to approve with conditions, 1 to resubmit for further review, and 1 against — approval followed the next day, July 19.⁴⁰ Medigen's measured results were a GMT ratio lower bound of 3.4 and a seroconversion-rate lower bound of 95.5%, both far above the threshold.³⁹

At ten in the morning on Sunday, August 15, the same standard was applied to measure United Biomedical. The venue was Meeting Room F327, third floor, Building F, of the TFDA's National Biotechnology Research Park; 22 people attended, the chair did not vote, and the outcome was 4 votes to resubmit for further review and 17 against. The meeting resolution read: "Given that the CDC has already determined that Taiwan's epidemic situation and vaccine needs constitute a genuine public health emergency, after evaluating the overall benefits and risks of the United Biomedical vaccine, its safety is acceptable and it shows a trend toward cellular immune response; however, compared with the external control group of Taiwanese recipients of the AZ vaccine, its neutralizing-antibody data failed to meet the efficacy-evaluation benchmark for COVID-19 vaccines that the expert meeting on May 28, 2021 established."³⁷

Source: TFDA meeting minutes, Vaccines (Basel) 2022;10(5):655, Liberty Times

📝 Curator's Note
The common claim is that "the standard was custom-tailored for Medigen." For that accusation to hold up, one thing needs explaining first: measured against the very same benchmark set on May 28 — the same one Medigen passed — United Biomedical failed on both counts. If the standard had really been calibrated for one company, a second company wouldn't have been stopped by the same ruler. The real technical disputes have nothing to do with Taiwan's blue-green political divide — they're about whether neutralizing antibodies, as a surrogate endpoint, can stand in for clinical protection, how comparable an external control group is to a head-to-head trial, and whether numbers measured against the original strain can be extrapolated to Delta. These were live debates across the international regulatory-science community in 2021, running on a track entirely parallel to Taiwan's political battles.

What makes those meeting minutes so compelling is the way each of the twenty-one experts who spoke stated their own position. Some wrote a single sentence; others filled a page with the details of animal studies. One expert was extremely specific about the data on participants 65 and older: "Among older (65 and up) participants, UB-612's seroconversion rate was only 89%, meaning that for every 9 people who received the United Biomedical vaccine, one failed to achieve even a basic immune response; combined with the fact that neutralizing-antibody GMT was lower among older participants than among the overall trial population, the United Biomedical vaccine's protection for the 65-and-older group is likely insufficient."³⁷

Another expert wrote, in that very same document: "Using immunogenicity (neutralizing antibodies) as a surrogate efficacy indicator has increasingly been discussed internationally."³⁷ That sentence undercuts the simplified narrative that "the review committee rejected immunobridging as a method." The committee voted against United Biomedical because its data didn't meet the threshold — not because of the method.

Immunobridging wasn't some sidedoor method Taiwan invented on its own, either. The International Coalition of Medicines Regulatory Authorities (ICMRA) held a workshop on June 24, 2021, co-chaired by Japan's PMDA and the UK's MHRA, with the FDA, EMA, and COVAX's advisory group all present; one of its conclusions was: "Immunogenicity bridging studies may be needed if an assessment of effectiveness of 2nd generation COVID-19 vaccines in clinical endpoint efficacy studies are no longer feasible."⁴¹

On September 15 that same year, the Access Consortium — made up of Australia's TGA, Health Canada, Singapore's HSA, Switzerland's Swissmedic, and the UK's MHRA — issued a statement: "Access Consortium members agree that well-justified and appropriately designed immunobridging studies are an acceptable approach for authorizing COVID-19 vaccines."⁴² Taiwan had set its own standard (on June 10) a full two weeks before that ICMRA workshop even took place.

The dissenting voices were just as specific. Academia Sinica academician Chen Pei-jer resigned from the TFDA's vaccine-efficacy expert review panel in late May, giving his reason as: "The biggest difficulty for the review is President Tsai — she has already said she'll get the domestic vaccine at the end of July. Under these circumstances, can the TFDA hold up against that pressure from the Presidential Office? It's a very simple point." He also said, "The review committee faces a great many difficulties; it can no longer maintain its independence and professionalism."⁴³ His concerns were about the political context surrounding the review's independence, not about the scientific basis of immunobridging itself.

Physician Lin Chin-chia's criticism targeted the method itself: "Medigen's 'immunobridging' approach is just a comparison between the Medigen vaccine group and the AZ vaccine group of neutralizing-antibody titers against live virus of the 'original strain' — this is nothing more than 'reinventing the wheel behind closed doors' in a laboratory." He also cited Germany's CureVac, whose Phase 3 trial of 40,000 participants showed only 47% efficacy, to illustrate that good immunogenicity doesn't equal high protection.⁴⁴ Some on the supporting side put it just as bluntly. Chang Hung-jen, former Deputy Director-General of the Department of Health, said: "The golden age of Phase 3 clinical trials for COVID-19 vaccines has already passed. Using traditional clinical trials to test the efficacy of next-generation vaccines will become increasingly unfeasible — immunobridging is bound to become vaccines' 'new Phase 3 trial.'"⁴⁵

Medigen did eventually run a Phase 3 trial, in Paraguay, in partnership with the medical school of the National University of Asunción; a 2022 interim analysis showed neutralizing-antibody titers reaching 3.7 times the AZ control group, and Paraguay's drug regulator, DINAVISA, approved an EUA on February 14, 2022. The sample size was only about 1,000 people, far smaller than the tens of thousands typical of a COVID vaccine Phase 3 trial, and the design was still immunobridging.⁴⁶ It happened, but its scale and design remained the focus of criticism.

August 23, 2021. President Tsai Ing-wen receives a dose of the domestically made Medigen vaccine. Photo: Makoto Lin / Office of the President, ROC (Taiwan) / Wikimedia Commons, CC BY 2.0

Of the 5 million Medigen doses the CECC procured, about 3.4 million were actually used, and roughly 1.6 million were destroyed after expiring.⁴⁷ In March 2026, the company obtained marketing approval in Vietnam for an Enterovirus 71 vaccine; its first shipment has already arrived in Vietnam, and it is simultaneously advancing development of a second-generation, multivalent enterovirus vaccine.⁴⁸ The same name that sparked street protests five years ago now makes headlines for exporting children's vaccines.

This fight burned from June through August, consuming the entire summer. It wasn't until fall that shipments really began arriving in volume — later than the fight ended, later than the virus had entered the community, and later than the time the people who needed it most could actually afford to wait.

The Shipments Arrived; Senior Citizens' Third Doses Still Lagged Behind

The memory of 2022 is two lines on a rapid test. Entire classes tested positive one after another; a few desks at the office emptied out every week; a few old neighbors in the village passed away. Testing positive went from being a major event to simply being your turn. Starting May 12, a rapid-test-positive result confirmed by a medical professional counted as a confirmed case for anyone already under home isolation or quarantine; this was expanded to everyone on May 26.⁴⁹ The single-day case peak was 94,808, announced by the CDC on May 27.⁵⁰ (Many retellings misremember the peak as occurring in June — that was a projection made in late April; the actual peak arrived earlier than expected.)

Premier Su Tseng-chang proposed the "New Taiwan Model" in April, with official goals set as zero deaths, a shift toward mild illness, and keeping the economy running normally.⁵¹ On October 13, entry quarantine rules changed to "0+7," waiving home quarantine entirely.⁵² Policy had turned decisively toward coexistence with the virus.

Supply was no longer the problem: by the end of August 2022, the government's cumulative procurement had reached 68.41 million doses, with NT$43.375 billion spent, including 41.05 million doses of Moderna and 7.6 million doses of BNT purchased directly by the government.⁵³

2022. Free at-home rapid test kits that National Health Insurance provided to people 65 and older. That year, two lines became an everyday verdict. Photo: ThomasYehYeh / Wikimedia Commons, CC BY-SA 4.0

What was missing was the final stretch. When the wave of deaths hit, the third-dose vaccination rate among people 75 and older still lagged behind their own first and second doses.⁶ Two years earlier, when doses couldn't be secured, the gap was at the docks. Two years later, with doses piled up in warehouses, the gap had moved to the last few miles: the front door of an elderly person living alone, a consent form at a long-term care facility, a booking phone call that no one was there to help make.

The tally for that year reads like this: Ministry of Health and Welfare mortality statistics show 14,667 COVID-19 deaths in 2022, with COVID-19's rank among causes of death jumping from 19th in 2021 to 3rd; the age distribution was concentrated among people 65 and older, at 86%, with a median age of death of 82 — ten years older than the year before.³ The Reporter used a different set of figures: 14,699 reported local deaths, with fatalities "concentrated almost entirely among the population 70 and older, at about 77%."⁵⁴ Two numbers, two different cutoffs, each internally consistent.

Chan Chang-chuan, a professor at National Taiwan University's College of Public Health, calculated the pace of that period: from May to December 2022, at least 25 people were confirmed positive every minute on average, and 5 people died of COVID every two hours on average.⁵⁴

Source: Ministry of the Interior household registration statistics, Ministry of Health and Welfare mortality statistics, The Reporter

Total nationwide deaths in 2022 came to 207,230, up 23,498 from 183,732 in 2021 — the first time the figure had topped 200,000 in the Ministry of the Interior's forty-six years of record-keeping.⁵⁵ There's another way to calculate "how many more people died that year," called excess mortality, and public health researchers have produced figures four to five times apart depending on which baseline they use.⁵⁶⁵⁷⁵⁸

Source: Taiwan Journal of Public Health 2024;43(2), Heho Health citing Hsueh Cheng-tai, Commercial Times citing Chen Yi-hua's team

📝 Curator's Note
The common claim is "Taiwan's death rate was the lowest in the world, so the cost was small." That sentence quietly swaps two different statistics into one. Cumulative mortality rate has population or confirmed cases as its denominator; it needs no counterfactual baseline, and it was pulled down by the extremely low base from 2020–2021. Excess mortality looks at the gap between a single year and a pre-pandemic baseline, and it needs a model. It's like a person who was healthy for two years and then came down with a bad flu in the third: "average sick days over three years was low" and "this year felt awful" can both be true at once. The cumulative mortality rate the CECC announced in February 2023 was 742.2 per million people, the sixth-lowest in a comparison of 38 OECD countries plus Taiwan and Singapore.⁵⁹ In October 2022, CECC commander Wang Pi-sheng's figure was that excess mortality was the third-lowest in the world.⁶⁰ Both were official announcements — but the set of countries being compared was different each time.

The eighteen months that were bought were not, in the end, fully spent on the people who needed them most. There's no conspiracy in that sentence, and it doesn't require anyone to have been derelict for it to be true — it's simply written into the median age of death.

Nearly Nine in Ten Applicants Received a Single Word: "Unrelated"

The cost didn't fall only on people who missed that one shot. People whose bodies reacted badly after getting the shot walked a different path altogether.

"I wasn't consulted at any point during the whole review process, and my medical history is very straightforward — how can they conclude, just from my health-checkup records, that the vaccine had nothing to do with my heart attack?" The person who said this is Chen Shih-chin, 51, who suffered a heart attack after receiving the AZ vaccine.⁶¹

On paper, the system isn't strict at all. A written report the Ministry of Health and Welfare submitted to the Legislative Yuan states plainly: under Article 6 of the Regulations on the Collection and Review of the Vaccination Injury Compensation Fund, members of the public need only submit an application — they bear no additional burden of proof, which instead falls on the administrative agency — and causation is determined primarily with reference to the World Health Organization's 2018-updated guidelines for causality assessment of adverse events following immunization.⁶²

A paper comparing the compensation systems of South Korea, Japan, and Taiwan grouped Taiwan together with Japan, describing both places' causation standards as "relatively lenient" — Taiwan's approach pays out compensation whenever causation is determined to be either "related" or "indeterminate."⁶³

"Compensation even when causation is indeterminate" is a design unique to Taiwan. As of a snapshot on February 17, 2022, there had been 4,485 cumulative applications, of which 351 had completed review: 47 received compensation (18 ruled "related," 29 ruled "indeterminate"), while 304 were ruled "unrelated" — nearly 87% ruled unrelated, with the review-completion rate under 8% of all applications.⁶⁴ Even with Taiwan compensating "indeterminate" cases, nearly nine in ten completed cases were still ruled unrelated to the vaccine.

Source: The Reporter, citing Vaccine Injury Compensation Program review records, as of 2022.02.17

By June 8, 2023, 2,689 cases had completed review, of which 303 were compensated (11.26%) and 2,386 were denied.⁶⁵

The review side has its own account, too. Chiu Nan-chang, a pediatric infectious disease physician at Mackay Memorial Hospital and convener of the review panel, explained the logic behind these rulings: "Data from around the world shows that mRNA vaccines can cause myocarditis, occurring mostly in young people after their second dose… but cancer and coronary artery blockage both develop over years, not as sudden events, so it doesn't make sense to attribute them to the vaccine."⁶⁶

The source of the money is also squeezing this system. Taiwan's compensation fund is financed by a NT$1.5 levy on every dose of vaccine administered, whereas Japan and South Korea fund theirs directly from the national treasury.⁶³ Lin Hsin-jou, an associate professor at Chang Gung University and a member of the review panel, used a metaphor: "Say the compensation fund was originally meant to handle measles vaccines, but now COVID-19 vaccines have been added into it… that shrinks the size of the original pool!"⁶¹

Compensation and subsidy payouts for COVID-19 vaccines reached a cumulative NT$153,025,000 by September 2023, already surpassing the NT$129,625,000 total paid out for all other vaccines combined over the thirty years from 1989 to 2020. In 2023, an average of 235 cases were reviewed per month — nearly eighteen times the 13 cases reviewed per month in 2020.⁶⁷ The single largest payout in one case was NT$6 million, to a 47-year-old woman who suffered cerebellar hemorrhage combined with cerebral venous sinus thrombosis and severe thrombocytopenia after receiving the AZ vaccine.⁶⁴

The second group bearing the cost was inside the hospitals. The nursing vacancy rate rose from 4.52% in fiscal year 2019 to 4.7% in 2021 and 6.53% in 2022; the turnover rate went from 11.12% to 10.13% and then to 11.73%.⁶⁸ In 2023, after the pandemic, the turnover rate hit 12.61% and the vacancy rate 9.05% — both the highest in nearly a decade.⁶⁹

On May 25, 2022, the CECC held a forum called "Listening to Nurses' Voices." Anonymous remarks from the floor included lines like: "If you think you can do it, come do it yourself," and "Would another NT$2,000 really buy us an extra pair of hands?" One nursing manager at a medical center demanded that the CECC specify a formula for how much general medical capacity should be cut for every increase in dedicated COVID wards, explaining: "Otherwise the hospital director just says to open more beds and we open them. We're just lowly nursing supervisors — how could we possibly tell the director that the nurses can't carry this much workload?"⁷⁰

Kao Ching-chiu, former chair of the National Union of Nurses Associations, took a longer view: "Between the falling birth rate and an industry environment that hasn't improved, the workforce isn't coming back — a shortage of staff may just become the new normal."⁶⁹

Mandated nurse-to-patient ratios across all three shifts were first implemented on March 1, 2024, through an incentive-based system; the Legislative Yuan passed an amendment to the Medical Care Act formally writing it into law on May 8, 2026, to be phased in starting May 20, 2027.⁷¹ Four years passed between that forum and the law taking effect.

The third group didn't even get to choose whether they were exposed — that was assigned to them. On June 2, 2021, a cluster broke out at the King Yuan Electronics plant in Miaoli; the large majority of the 500-plus confirmed cases were migrant workers. On June 7, the Miaoli County Government announced a ban on all foreign migrant workers countywide leaving their residences, with no exceptions besides commuting to and from work; the ban was lifted on June 29.⁷² Taiwanese employees at the very same plant faced no restriction of comparable severity.

The CECC's own legal opinion at the time was already unambiguous: the ban "raised concerns of differential treatment in violation of the principle of equality," and under the intent of the Constitution, international human rights covenants, and the Communicable Disease Control Act, it "should carry no binding legal force."⁷³ The Taiwan Association for Human Rights sent a letter to central government agencies on June 18 demanding an explanation. The ban nonetheless remained in effect until June 29.

A year later, the Control Yuan issued a formal corrective measure against both the Miaoli County Government and the Ministry of Labor. The corrective measure stated that "the Miaoli County Government issued a lockdown order restricting all migrant workers from leaving home — including those with no history of exposure to the disease — without clear legal authorization," and noted that the CECC had failed to clearly demand its immediate withdrawal, which "gave rise to discriminatory treatment of migrant workers and constitutes a major dereliction of duty." This incident was written into the annual human rights reports of both Amnesty International and the U.S. State Department.⁷⁴

✦ "I felt like I was some kind of criminal."

The person who said this was Emma, an Indonesian migrant worker, in an interview with The Reporter on June 22, 2021. She also said: "We're afraid of getting infected too, you know."⁷⁵

The losses in insurance and dining were on a different scale entirely. The paired "pandemic" and "vaccine-injury" insurance policies together paid out a combined NT$264.983 billion in claims from 2022 through the end of March 2023 — wiping out roughly two decades of the property-and-casualty insurance industry's profits.⁷⁶ In May 2021, the month of the nationwide Level 3 alert and the ban on dine-in service, restaurant industry revenue fell to NT$50.2 billion, down 23.7% month-on-month and 19.1% year-on-year; everyday-dining restaurants fell 30.8% year-on-year, the steepest drop of any category, while Western-style fast food actually grew 5.4%, thanks to its more takeout-oriented customer base.⁷⁷

There are still other groups who were never counted separately, from beginning to end. The number of deaths that year among long-term care facility residents and elderly people living alone was never published by any single agency on its own. There is no official survey of how delivery riders' and day-wage workers' incomes changed during the Level 3 alert. There are no figures for migrant workers' vaccination coverage, or for wage compensation after those three weeks of lockdown. There is no public tally of how many people pursued an appeal or administrative litigation after being ruled "unrelated." These groups' circumstances have never been entered into any table at all.

No Illegality Found; Not Yet Thorough Enough; Causation Not Established

After it was all over, three different agencies each conducted a review and arrived at three seemingly contradictory conclusions.

The Taipei District Prosecutors Office accepted more than sixty vaccine-related complaints between June 2021 and February 2023, covering allegations ranging from inadequate procurement and refusal of private donations to the Medigen procurement policy and pricing and the special-approval process. In April 2023, it issued a press release stating that all of them had been closed with a finding of "no illegality found."⁷⁸

The Control Yuan released its investigative report on vaccine procurement on March 1, 2023, demanding that the Executive Yuan direct the Ministry of Health and Welfare to review and improve areas where vaccine preparedness, procurement decisions, and operating procedures had "not yet reached full adequacy," along with incomplete relevant regulations and insufficient information disclosure. One item concerned the limited international recognition of the domestic vaccine: Control Yuan members noted that, at the time, only ten countries recognized the Medigen vaccine internationally, and the main reason was likely the absence of traditional Phase 3 clinical trial results demonstrating efficacy.⁷⁹

On the courts' side, a two-year-old boy died after contracting the virus, and his father, questioning delays in the emergency-response process, sought NT$1 in state compensation from the New Taipei City Government. The court ruled that it was "difficult to establish substantial causation with the actions of public officials," and the second-instance ruling in November 2023 rejected the appeal, finalizing the case.⁸⁰

The three agencies were asking three different questions. The prosecutors asked whether anyone had committed a criminal offense. The Control Yuan asked whether administrative procedures had been adequate and information disclosure sufficient. The courts asked whether there was substantial causation between a specific death and a specific official's actions. "Illegality," "inadequacy," and "no substantial causation" are three different legal standards; each of the three tracks answered its own question, and each left its own share of questions unanswered. Reading them as contradicting one another means mistaking three different rulers for a single one.

Scan the QR code before entering, send the text message to 1922. An action Taiwanese people performed billions of times — not one record of it exists anymore. Photo: IiTheCavalryiI / Wikimedia Commons, CC BY-SA 4.0

The one thing none of the three rulers could measure is the same thing: in a state of emergency, who sets the rules, and when.

The first rule concerns whether decisions leave a paper trail. The April 13, 2021 meeting left no minutes, and after its investigation, the Control Yuan acknowledged that existing regulations didn't actually require any — which is exactly why it recommended amending the law to specify whether major epidemic-prevention decisions should be documented and made public.¹⁸ This rule wasn't proposed until after the fact.

The second rule concerns whether the scope of authorization was written clearly enough. Article 7 of the Special Act for Prevention, Relief and Revitalization Measures for Severe Pneumonia with Novel Pathogens, passed on its third reading on February 25, 2020, consists of a single sentence in its entirety: "The commander of the Central Epidemic Command Center may implement necessary emergency response measures as needed to prevent and control the epidemic." This act remained in effect until it automatically expired on June 30, 2023.⁸¹

Lin Ming-shin, a professor at National Taiwan University's College of Law, wrote a dedicated study in a legal journal examining this article's constitutionality, approaching it from three angles — the constitutional mandate of clarity, the clarity of legislative authorization, and the separation of powers — and stressed that even though this article had not yet been abused in practice, the question of its constitutionality still warranted examination.⁸² The Control Yuan used a different vocabulary, noting that the current legal system "relies heavily on general clauses within the law" and that "turning ordinary law into emergency law" risked expanding executive power.¹¹ The scholar's terms differ from the Control Yuan's, but they're pointing at the same thing.

The third rule concerns how data is collected and deleted. Technology-driven epidemic prevention wasn't without a legal basis: the electronic fence cited Article 48, Paragraph 1 of the Communicable Disease Control Act — "the competent authority may detain for observation any person who has been in contact with, or is suspected of having been infected by, a communicable disease patient; and where necessary, may order such person to relocate to a designated location for examination, vaccination, medication, or other necessary measures such as designating control or isolation in a specific area" — together with Article 58, Paragraph 1, Item 4, and Article 20 of the Personal Data Protection Act.⁸³ None of these provisions ever mentions mobile phone numbers or cell tower positioning. The dispute is over whether the authorization was clearly defined, not over whether a legal basis existed at all.

The Control Yuan's investigation into the electronic fence found shortcomings across all four dimensions of procedure, administrative transparency, personal data protection, and legal framework: the system went live in early February 2020, but the official document formalizing its operating procedures wasn't issued until March 8, and the contract for the system's service platform didn't take effect until March 26.

High-risk groups in Wanhua District were placed under National Health Insurance card flagging starting May 15, 2021, yet this wasn't announced at a press conference until May 28 — just as the flagging period was about to end. The "maximum 28 days" limit on data retention wasn't put in writing until four months after the system had already gone live, and telecom carriers separately operated under a one-year retention rule.¹¹

The rules for the SMS-based real-name contact-tracing system were likewise made up as they went. From its launch on May 19, 2021, to the system going offline on April 27, 2022, the National Communications Commission reported nearly 4.8 billion messages sent in total; the data was completely deleted on May 26, 2022.⁸⁴

Chang Yuan-sen, a judge at the Taichung District Court, pointed out while this system was still in operation that police using contact-tracing SMS data as grounds for search-warrant applications was troubling — members of the public might have been compelled to participate simply to buy daily necessities, and criminal investigations fell outside the authorized scope of "for epidemiological investigation use only." The CECC responded that the data could not be used for criminal investigations; the Criminal Investigation Bureau clarified that it had not directly accessed the Ministry of Health and Welfare's 1922 SMS data but had instead obtained the full message contents through the procedures of the Communication Security and Surveillance Act; and the Judicial Yuan stated that the matter should either be judged case by case by individual judges or clearly regulated through more precise legislation.⁸⁵

💡 Did You Know
Taiwan's contact tracing followed the path of cell tower positioning plus SMS reporting, while many other countries in the same period chose decentralized Bluetooth-based exposure notification — two phones exchange random codes when they come near each other, and the codes stay on the phone rather than uploading any location. The privacy costs of the two paths are very different. Why Taiwan chose the former was never part of any public debate at the time; what left behind a large paper trail was the legal controversy that came later, not the choice of technical architecture.

The CECC wound itself down step by step in 2023. Indoor mask requirements ended on February 20; on March 20, mild cases no longer had to be reported and shifted to self-health management instead; on May 1, COVID-19 was downgraded from a Category 5 to a Category 4 notifiable disease, and the CECC disbanded that same day.⁸⁶ The press conference held on April 27 was its 960th, and its last. Over three years, it had held 960 press conferences, issued 2,199 press releases, produced nearly seven thousand briefing boards, and drawn more than twenty thousand reporter attendances in total, with as many as 220,000 people in quarantine at the same time at its peak.⁸⁷

Over more than three years, more than ten million people in Taiwan were infected in total. The cumulative death toll doesn't reconcile across different official figures: on the eve of the CECC's disbandment, the Central News Agency cited a figure of more than 19,000⁸⁷, while a snapshot from an open data platform that September gave cumulative confirmed cases as 10,239,510 and cumulative deaths as 17,668.⁸⁸

PTS News report from April 27, 2023: the Central Epidemic Command Center's 960th and final press conference.

On the stage beyond its own shores, Taiwan's situation was a different story altogether: it was invited to attend the World Health Assembly as an observer under the name "Chinese Taipei" from 2009 to 2016, and has not been invited again since 2017.

Then-Minister of Health and Welfare Hsueh Jui-yuan said in Geneva in 2023: "Without WHO membership, Taiwan is also unable to provide various surveillance data to the global influenza surveillance and the response system, which could alert the world to the next pandemic. Taiwan is willing and it should also be included in the pandemic accord that is under negotiation."⁸⁹

On May 27, 2025, the 78th World Health Assembly adopted the Pandemic Agreement and the Pathogen Access and Benefit Sharing (PABS) system; eleven of Taiwan's diplomatic allies proposed inviting Taiwan to attend as an observer, but Taiwan still was not invited.⁹⁰

The world did write Taiwan's pandemic response into its academic record. That European public health journal article also noted that Taiwan avoided a nationwide lockdown in favor of targeted restrictions, while also pointing out that the vulnerability of its elderly population led to higher mortality and case rates, and that it lacked a clear transition plan.⁴ During that same period, Taiwan's surveillance data still couldn't get into the global system.

On August 4, 2026, the CDC announced the previous week's numbers: 18,990 outpatient and emergency-room visits for COVID, up 74% from the week before, along with 62 new severe local cases and 3 new local deaths. Director-General Lo Yi-chun estimated the peak would land around August 16, with that week's visits potentially reaching 51,000.⁹¹ The person who, seven years earlier, saw the reports of an unexplained pneumonia in Wuhan on that early morning was then deputy director-general and spokesperson of the CDC. He is now its director-general. The same virus, the same person, seven years apart.

In those eighteen months, Taiwan really did buy time. Whether that time was ultimately spent on the people who needed it most is a separate question — and the answer is written into the median age of death.

The threads still left hanging are just as clear. The Tzu Chi fraud case was indicted on August 6, 2026, and has not yet gone to trial. No real-world effectiveness studies following the immunobridging approvals can be found in the public literature. There is no public tally of appeals or administrative litigation pursued after a ruling of "unrelated." No agency has left behind figures on the situation of long-term care residents, elderly people living alone, or migrant workers over those three years. Publicly funded vaccination is still ongoing in 2026, and outpatient visit numbers are still climbing. These are the facts as they stood as of August 8, 2026.

Further Reading

• Taiwan's Public Health and Epidemic Prevention System — How the trauma of SARS grew into an entire system, and what that system looked like before 2020
• Taiwan's Healthcare and National Health Insurance — The health insurance card that let the mask rationing system stand up within two weeks, and the system underneath it
• Migrant Workers — The structural background behind the Miaoli lockdown order: the labor and living conditions of 700,000 people in Taiwan
• Audrey Tang — The digital minister in that mask-map coordination chain, and the relationship between civic tech and open government data
• Taiwan's Disaster Medicine System — The institutional background behind dedicated COVID wards and emergency-room congestion
• Academia Sinica — The institution that produced a monoclonal antibody able to identify the COVID-19 virus in nineteen days, and the system that let it move that fast

Image Sources

This article uses 7 images (1 CC0, 6 Creative Commons licensed), all cached at public/article-images/society/ to avoid hotlinking source servers; it also embeds 2 videos from PTS News's official channel. Neither video's captions were extracted, so the body text does not quote anything said in either video.

• Large-scale COVID-19 vaccination site at Taipei Expo Dome — Photo: 玄史生, 2021-10-27, CC0 1.0 (hero)
• Lining up to buy rationed masks outside a pharmacy in Hsinchu City — Photo: Tbatb, 2020-03-30, CC BY-SA 4.0
• Taipei MRT's Dongmen Station during the Level 3 alert — Photo: 稜光映像_Brian, 2021-05-25, CC BY 3.0
• Intersection signage outside a rapid-testing station in Zhonghe District, New Taipei City — Photo: Tsuna Lu (Adsa562), 2021-06-12, CC BY-SA 4.0
• President Tsai Ing-wen receives a dose of the domestically made Medigen vaccine — Photo: Makoto Lin / Office of the President, ROC (Taiwan), 2021-08-23, CC BY 2.0
• Free at-home rapid test kits provided by National Health Insurance to people 65 and older — Photo: ThomasYehYeh, 2022-11-07, CC BY-SA 4.0
• SMS-based contact-tracing QR code notice at a shop entrance — Photo: IiTheCavalryiI, 2022-05-23, CC BY-SA 4.0
• Videos: PTS News, "Recalibration" report, 2021-05-22; PTS News, "CECC's Final Press Conference" report, 2023-04-27 (official YouTube channel)

References